I bookmarked this thread, wanting to reply earlier... but um... I was busy working on organized play for D&DXP (hey, at least I have a good excuse!).
The major "living" campaigns we have seen from WotC and others have hardly been "living" in any sense.
The concept may be something that is too difficult to even tackle. I believe that's why the nomenclature has gone away from using the word "living" in some of the newer campaigns (Ashes of Athas, for example).
As Shawn said, the term "living" can mean various things. For AoA, I'm fairly sure we (I'm one of the admins) avoided the term because initially the campaign was going to be only at conventions. We also wanted to avoid intruding on LFR's space for various reasons. Dropping the term was a way of being honest about the campaign having limits and not being WotC's main campaign.
That said, we write it around a living framework. It isn't LG by any means (LG had armies of volunteers, active forums with LARP-style interactions, many interactives, and tons of ways for players and campaign staff to interact). But, Ashes does record what you do and we absolutely respond to what players do and make huge changes.
For example, the decision to have the True be soundly defeated (at least, so it seems... ahem) and to start Chapter 4 on a new track was based in part on how well PCs did in AOA3-3 and the read we had from players. It felt like the end of a story arc, so the True have faded into the background a bit. The heroes have also made a number of decisions that have influenced various NPCs (both good and nefarious) that will change future events (and already has in some cases). But, these things are not immediate. A player could even say they don't feel them at all.
To my knowledge, the stuff I discussed in the prior paragraph is not something anyone has attempted to tackle. Updating prior (or future) adventures based on the actions of players is a monumental, possibly Herculean task.
It is. One of the keys is that you need to write the next series of adventures often before you can have post-convention play for the previous adventures you had written. It takes 3-4 months to write several high quality adventures. In theory it could be less, but it is extremely hard to do so.
You can have trigger points in your plot, such as:
If the town is destroyed, the next adventure is in a different town.
If the town is save, the next adventure is in this town.
But those changes are less significant than saying you have an entirely different adventure (either about going after and routing the attackers or about trying to rebuild from destruction). It is very hard to react quickly and introduce significant change.
Also, players react negatively when they perceive punishment. "You lost the interactive, so in this adventure your queen is dead and you can't really get a feelgood ending." Because of that, it is hard to have "teeth" from results. It is possible, just more challenging.
One way to accomplish this would involve tabulating surveys (or something to that effect) from the end of each adventure played and somehow work the results into some sort of update on a monthly or annual basis to change the setting and past/future adventures to reflect what has been done in the "living" part of the campaign.
I think this is the most realistic way to do so. If you can have almost a parallel success track to the story track, and have it be visible and measurable that would work well. For example, the campaign revolves around a frontier town and each scenario has success/fail points, which in turn affect the growth of the town and resources available. By keeping that track separate you could still write mods in advance, with plugins based on success levels.
Alternately, the success could be based on the table, adding up each players' success points.
Both are "gamist" approaches in that they are obvious systems rather than subtle story.